Cytochrome P450 Drives a HIF-Regulated Behavioral Response to Reoxygenation by C. elegans Dengke K. Ma, 1 Michael Rothe, 2 Shu Zheng, 1 Nikhil Bhatla, 1 Corinne L. Pender, 1 Ralph Menzel, 3 H. Robert Horvitz 1 * Oxygen deprivation followed by reoxygenation causes pathological responses in many disorders, including ischemic stroke, heart attacks, and reperfusion injury. Key aspects of ischemia-reperfusion can be modeled by a Caenorhabditis elegans behavior, the O2-ON response, which is suppressed by hypoxic preconditioning or inactivation of the O 2 -sensing HIF (hypoxia-inducible factor) hydroxylase EGL-9. From a genetic screen, we found that the cytochrome P450 oxygenase CYP-13A12 acts in response to the EGL-9–HIF-1 pathway to facilitate the O2-ON response. CYP-13A12 promotes oxidation of polyunsaturated fatty acids into eicosanoids, signaling molecules that can strongly affect inflammatory pain and ischemia-reperfusion injury responses in mammals. We propose that roles of the EGL-9–HIF-1 pathway and cytochrome P450 in controlling responses to reoxygenation after anoxia are evolutionarily conserved. I schemia-reperfusion–related disorders, such as strokes and heart attacks, are the most common causes of adult deaths worldwide (1). Blood delivers O 2 and nutrients to target tis- sues, and ischemia results when the blood supply is interrupted. The restoration of O 2 from blood flow after ischemia, known as reperfusion, can exacerbate tissue damage (2). How organisms prevent ischemia-reperfusion injury is poorly un- derstood. Studies of the nematode C. elegans led to the discovery of an evolutionarily conserved fam- ily of O 2 -dependent enzymes (EGL-9 in C. elegans and EGLN2 in mammals) that hydroxylate the HIF transcription factor and link hypoxia to hypoxia- inducible factor (HIF)–mediated physiological responses (3–7). Exposure to chronic low con- centrations of O 2 (hypoxic preconditioning) or direct inhibition of EGLN2 strongly protects mam- mals from stroke and ischemia-reperfusion in- jury (2, 8, 9). Similarly, EGL-9 inactivation in C. elegans blocks a behavioral response to reoxy- genation, the O2-ON response (characterized by a rapidly increased locomotion speed triggered by reoxygenation after anoxia) ( 10, 11), which is similar to mammalian tissue responses to ischemia- reperfusion: (i) Reoxygenation drives the O2- ON response and is the major pathological driver of reperfusion injury, (ii) hypoxic preconditioning can suppress both processes, and (iii) the central regulators (EGL-9–HIF) of both processes are evolutionarily conserved. It remains unknown how the EGL-9–HIF-1 and EGLN2-HIF path- ways control the O2-ON response and ischemia- reperfusion injury, respectively. To seek EGL-9–HIF-1 effectors important in the O2-ON response, we performed an egl-9 suppressor screen for mutations that can restore the defective O2-ON response in egl-9 mutants (fig. S1A). We identified new alleles of hif-1 in this screen; because EGL-9 inhibits HIF-1, hif-1 mutations suppress the effects of egl-9 muta- tions (10). We also identified mutations that are not alleles of hif-1 (Fig. 1, A to C, and fig. S1B). hif-1 mutations recessively suppressed three defects of egl-9 mutants: the defective O2-ON response, defects in egg laying, and the ectopic expression of the HIF-1 target gene cysl-2 (pre- viously called K10H10.2) (fig. S1C) (10, 12). By contrast, one mutation, n5590, dominantly sup- pressed the O2-ON defect but did not suppress the egg-laying defect or the ectopic expression of cysl-2::GFP (Fig. 1, D and E, and fig. S2). n5590 restored the sustained phase (starting 30 s after reoxygenation) better than it did the initial phase (within 30 s after reoxygenation) (Fig. 1, A to C). egl-9; hif-1; n5590 triple mutants dis- played a normal O2-ON response, just like the wild type and egl-9; hif-1 double mutants (fig. S1D). Thus, n5590 specifically suppresses the egl-9 defect in the sustained phase of the O2-ON response. We genetically mapped n5590 and identified a Met 46 → Ile missense mutation in the gene cyp-13A12 by whole-genome sequencing (Fig. 2A, fig. S3A, and table S1A). Decreased wild- type cyp-13A12 gene dosage in animals heter- ozygous for a wild-type allele and the splice acceptor null mutation gk733685, which truncates the majority of the protein, did not recapitulate the dominant effect of n5590 (Fig. 2B). Similarly, gk733685 homozygous mutants did not recapit- ulate the effect of n5590 (Fig. 2C). Thus, n5590 does not cause a loss of gene function. By con- trast, increasing wild-type cyp-13A12 gene dos- age by overexpression restored the sustained phase of the O2-ON response (Fig. 2D), and RNA interference (RNAi) against cyp-13A12 abolished the effect of n5590 (Fig. 2E). We con- clude that n5590 is a gain-of-function allele of cyp-13A12. cyp-13A12 encodes a cytochrome P450 oxy- genase (CYP). CYPs can oxidize diverse sub- strates (13–15). The C. elegans genome contains about 82 CYP genes, at least two of which are polyunsaturated fatty acid (PUFA) oxygenases that generate eicosanoid signaling molecules (fig. S3B) (16, 17). On the basis of BLASTP scores, the closest human homolog of CYP-13A12 is CYP3A4 (fig. S4). We aligned the protein se- quences of CYP-13A12 and CYP3A4 and found that n5590 converts methionine 46 to an iso- leucine, the residue in the corresponding position of normal human CYP3A4 (fig. S4). Methio- nines can be oxidized by free radicals, which are produced in the CYP enzymatic cycle, rendering CYPs prone to degradation (18, 19). Using tran- scriptional and translational green fluorescent pro- tein (GFP)–based reporters, we identified the pharyngeal marginal cells as the major site of expression of cyp-13A12 (fig. S5) and observed that the abundance of CYP-13A12::GFP pro- tein was decreased by prolonged hypoxic pre- conditioning and also decreased in egl-9 but not in egl-9; hif-1 mutants (Fig. 2F and fig. S5). The n5590 mutation prevented the decrease in CYP-13A12::GFP abundance by hypoxia or egl-9. Thus, n5590 acts, at least in part, by restoring the normal abundance of CYP-13A12, which then promotes the O2-ON response in egl-9 mutants. We tested whether CYP-13A12 was normally required for the O2-ON response in wild-type animals. The cyp-13A12 null allele gk733685 abolished the sustained phase of the O2-ON re- sponse; the initial phase of the O2-ON response was unaffected (Fig. 3A). A wild-type cyp-13A12 transgene fully rescued this defect (Fig. 3B). A primary role of CYP-13A12 in the sustained phase of the O2-ON response explains the in- complete rescue of the defective O2-ON response of egl-9 mutants by n5590 during the initial phase (Fig. 1C). The activity of most and possibly all C. elegans CYPs requires EMB-8, a CYP re- ductase that transfers electrons to CYPs (20). No non-CYP EMB-8 targets are known. The mutation emb-8(hc69) causes a temperature-sensitive em- bryonic lethal phenotype. We grew emb-8(hc69) mutants at the permissive temperature to the young-adult stage. A shift to the nonpermissive temperature simultaneously with Escherichia coli–feeding RNAi against emb-8 nearly abol- ished the O2-ON response (Fig. 3, C and D). [Both the hc69 mutation and RNAi against emb- 8 were required to substantially reduce the level of EMB-8 (17).] CYP-13A12 is thus required for the sustained phase of the O2-ON response, and one or more other CYPs likely act with CYP- 13A12 to control both phases of the O2-ON response. CYP oxygenases define one of three enzyme families that can convert PUFAs to eicosanoids, which are signaling molecules that affect inflam- 1 Howard Hughes Medical Institute, Department of Biology, McGovern Institute for Brain Research, Koch Institute for Inte- grative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA. 2 Lipidomix GmbH, Robert-Roessle-Str. 10, 13125 Berlin, Germany. 3 Freshwater and Stress Ecology, Department of Biology, Humboldt-Universität zu Berlin, Spaethstr. 80/81, 12437 Berlin, Germany. *Corresponding author. E-mail: [email protected]2 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org 554 REPORTS
Transcript
Cytochrome P450 Drives aHIF-Regulated Behavioral Responseto Reoxygenation by C. elegansDengke K. Ma,1 Michael Rothe,2 Shu Zheng,1 Nikhil Bhatla,1 Corinne L. Pender,1Ralph Menzel,3 H. Robert Horvitz1*
Oxygen deprivation followed by reoxygenation causes pathological responses in many disorders,including ischemic stroke, heart attacks, and reperfusion injury. Key aspects of ischemia-reperfusioncan be modeled by a Caenorhabditis elegans behavior, the O2-ON response, which is suppressedby hypoxic preconditioning or inactivation of the O2-sensing HIF (hypoxia-inducible factor)hydroxylase EGL-9. From a genetic screen, we found that the cytochrome P450 oxygenaseCYP-13A12 acts in response to the EGL-9–HIF-1 pathway to facilitate the O2-ON response.CYP-13A12 promotes oxidation of polyunsaturated fatty acids into eicosanoids, signaling moleculesthat can strongly affect inflammatory pain and ischemia-reperfusion injury responses in mammals.We propose that roles of the EGL-9–HIF-1 pathway and cytochrome P450 in controllingresponses to reoxygenation after anoxia are evolutionarily conserved.
Ischemia-reperfusion–related disorders, suchas strokes and heart attacks, are the mostcommon causes of adult deaths worldwide
(1). Blood delivers O2 and nutrients to target tis-sues, and ischemia results when the blood supplyis interrupted. The restoration of O2 from bloodflow after ischemia, known as reperfusion, canexacerbate tissue damage (2). How organismsprevent ischemia-reperfusion injury is poorly un-derstood. Studies of the nematode C. elegans ledto the discovery of an evolutionarily conserved fam-ily of O2-dependent enzymes (EGL-9 inC. elegansand EGLN2 in mammals) that hydroxylate the HIFtranscription factor and link hypoxia to hypoxia-inducible factor (HIF)–mediated physiologicalresponses (3–7). Exposure to chronic low con-centrations of O2 (hypoxic preconditioning) ordirect inhibition of EGLN2 strongly protectsmam-mals from stroke and ischemia-reperfusion in-jury (2, 8, 9). Similarly, EGL-9 inactivation inC. elegans blocks a behavioral response to reoxy-genation, theO2-ON response (characterized bya rapidly increased locomotion speed triggeredby reoxygenation after anoxia) (10, 11), which issimilar to mammalian tissue responses to ischemia-reperfusion: (i) Reoxygenation drives the O2-ON response and is the major pathological driverof reperfusion injury, (ii) hypoxic preconditioningcan suppress both processes, and (iii) the centralregulators (EGL-9–HIF) of both processes areevolutionarily conserved. It remains unknownhow the EGL-9–HIF-1 and EGLN2-HIF path-ways control the O2-ON response and ischemia-reperfusion injury, respectively.
To seek EGL-9–HIF-1 effectors importantin the O2-ON response, we performed an egl-9suppressor screen for mutations that can restorethe defective O2-ON response in egl-9mutants(fig. S1A). We identified new alleles of hif-1 inthis screen; because EGL-9 inhibits HIF-1, hif-1mutations suppress the effects of egl-9 muta-tions (10). We also identified mutations that arenot alleles of hif-1 (Fig. 1, A to C, and fig. S1B).hif-1 mutations recessively suppressed threedefects of egl-9 mutants: the defective O2-ONresponse, defects in egg laying, and the ectopicexpression of the HIF-1 target gene cysl-2 (pre-viously called K10H10.2) (fig. S1C) (10, 12). Bycontrast, one mutation, n5590, dominantly sup-pressed the O2-ON defect but did not suppressthe egg-laying defect or the ectopic expressionof cysl-2::GFP (Fig. 1, D and E, and fig. S2).n5590 restored the sustained phase (starting 30 safter reoxygenation) better than it did the initialphase (within 30 s after reoxygenation) (Fig. 1,A to C). egl-9; hif-1; n5590 triple mutants dis-played a normal O2-ON response, just like thewild type and egl-9; hif-1 double mutants (fig.S1D). Thus, n5590 specifically suppresses theegl-9 defect in the sustained phase of the O2-ONresponse.
We genetically mapped n5590 and identifieda Met46 → Ile missense mutation in the genecyp-13A12 by whole-genome sequencing (Fig.2A, fig. S3A, and table S1A). Decreased wild-type cyp-13A12 gene dosage in animals heter-ozygous for a wild-type allele and the spliceacceptor null mutation gk733685, which truncatesthe majority of the protein, did not recapitulatethe dominant effect of n5590 (Fig. 2B). Similarly,gk733685 homozygous mutants did not recapit-ulate the effect of n5590 (Fig. 2C). Thus, n5590does not cause a loss of gene function. By con-trast, increasing wild-type cyp-13A12 gene dos-age by overexpression restored the sustainedphase of the O2-ON response (Fig. 2D), and
RNA interference (RNAi) against cyp-13A12abolished the effect of n5590 (Fig. 2E). We con-clude that n5590 is a gain-of-function allele ofcyp-13A12.
cyp-13A12 encodes a cytochrome P450 oxy-genase (CYP). CYPs can oxidize diverse sub-strates (13–15). The C. elegans genome containsabout 82 CYP genes, at least two of which arepolyunsaturated fatty acid (PUFA) oxygenasesthat generate eicosanoid signaling molecules (fig.S3B) (16, 17). On the basis of BLASTP scores,the closest human homolog of CYP-13A12 isCYP3A4 (fig. S4). We aligned the protein se-quences of CYP-13A12 andCYP3A4 and foundthat n5590 converts methionine 46 to an iso-leucine, the residue in the corresponding positionof normal human CYP3A4 (fig. S4). Methio-nines can be oxidized by free radicals, which areproduced in the CYP enzymatic cycle, renderingCYPs prone to degradation (18, 19). Using tran-scriptional and translational green fluorescent pro-tein (GFP)–based reporters, we identified thepharyngeal marginal cells as the major site ofexpression of cyp-13A12 (fig. S5) and observedthat the abundance of CYP-13A12::GFP pro-tein was decreased by prolonged hypoxic pre-conditioning and also decreased in egl-9 butnot in egl-9; hif-1mutants (Fig. 2F and fig. S5).The n5590 mutation prevented the decrease inCYP-13A12::GFP abundance by hypoxia or egl-9.Thus, n5590 acts, at least in part, by restoringthe normal abundance of CYP-13A12, which thenpromotes the O2-ON response in egl-9 mutants.
We tested whether CYP-13A12was normallyrequired for the O2-ON response in wild-typeanimals. The cyp-13A12 null allele gk733685abolished the sustained phase of the O2-ON re-sponse; the initial phase of the O2-ON responsewas unaffected (Fig. 3A). Awild-type cyp-13A12transgene fully rescued this defect (Fig. 3B).A primary role of CYP-13A12 in the sustainedphase of the O2-ON response explains the in-complete rescue of the defective O2-ON responseof egl-9mutants by n5590 during the initial phase(Fig. 1C). The activity of most and possibly allC. elegans CYPs requires EMB-8, a CYP re-ductase that transfers electrons to CYPs (20). Nonon-CYP EMB-8 targets are known. The mutationemb-8(hc69) causes a temperature-sensitive em-bryonic lethal phenotype. We grew emb-8(hc69)mutants at the permissive temperature to theyoung-adult stage. A shift to the nonpermissivetemperature simultaneously with Escherichiacoli–feeding RNAi against emb-8 nearly abol-ished the O2-ON response (Fig. 3, C and D).[Both the hc69mutation and RNAi against emb-8 were required to substantially reduce the levelof EMB-8 (17).] CYP-13A12 is thus required forthe sustained phase of the O2-ON response, andone or more other CYPs likely act with CYP-13A12 to control both phases of the O2-ONresponse.
CYP oxygenases define one of three enzymefamilies that can convert PUFAs to eicosanoids,which are signaling molecules that affect inflam-
1Howard Hughes Medical Institute, Department of Biology,McGovern Institute for Brain Research, Koch Institute for Inte-grative Cancer Research,Massachusetts Institute of Technology,Cambridge,MA02139,USA. 2Lipidomix GmbH, Robert-Roessle-Str.10, 13125 Berlin, Germany. 3Freshwater and Stress Ecology,Department of Biology, Humboldt-Universität zu Berlin, Spaethstr.80/81, 12437 Berlin, Germany.
2 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org554
REPORTS
matory pain and ischemia-reperfusion responsesof mammals (15, 21–23); the other two families,cyclooxygenases and lipoxygenases, do not ap-pear to be present in C. elegans (17, 24). To testwhether eicosanoids are regulated by EGL-9 andCYP-13A12, we used high-performance liquidchromatography (HPLC) coupled withmass spec-trometry (MS) to profile steady-state amountsof 21 endogenous eicosanoid species from cellextracts of wild-type, egl-9(n586), and egl-9(n586);cyp-13A12(n5590) strains. Only free eicosanoidshave potential signaling roles (21, 22, 24), so wefocused on free eicosanoids. The egl-9 mutationcaused a marked decrease in the overall amountof free eicosanoids, whereas the total amount ofeicosanoids, including both free and membrane-bound fractions, was unaltered (Fig. 4A and fig.S6).Among the eicosanoids profiled, 17,18-DiHEQ(17,18-diolhydroxyeicosatetraenoic acid) was themost abundant species (fig. S6B). 17,18-DiHEQis the catabolic hydrolase product of 17,18-EEQ(17,18-epoxyeicosatetraenoic acid), an epoxideactive in eicosanoid signaling (25). Free cytosolic
17,18-EEQ and 19-hydroxyeicosatetraenoic acid(19-HETE) were present in the wild type butundetectable in egl-9 mutants (Fig. 4, C to F).egl-9(n586); cyp-13A12(n5590) mutants exhibitedpartially restored free overall eicosanoid levels aswell as restored levels of 17,18-EEQ and 19-HETE(Fig. 4, A to F, and fig. S6B). Thus, both EGL-9and CYP-13A12 regulate amounts of free cyto-solic eicosanoids.
We tested whether the O2-ON response re-quires PUFAs, which are CYP substrates andeicosanoid precursors. PUFA-deficient fat-2 andfat-3 mutants (26) exhibited a complete lack ofthe O2-ON response, although the accelerationin response to anoxia preceding the O2-ON re-sponse was normal (Fig. 4G and fig. S7, A to C).The defective O2-ON response of fat-2 mutantswas restored by feeding animals arachidonicacid, a C20 PUFA (Fig. 4H), but not oleate, aC18 monounsaturated fatty acid that is processedby FAT-2 to generate C20 PUFAs (fig. S7D).These results demonstrate an essential role ofPUFAs for the O2-ON response.
We suggest a model in which CYPs, whichare strictly O2-dependent (27, 28), generate ei-cosanoids to drive the O2-ON response (Fig. 4Iand fig. S8). In this model, EGL-9 acts as achronic O2 sensor, so that during hypoxic pre-conditioning, the O2-dependent activity of EGL-9is inhibited, HIF-1 is activated, and unknownHIF-1 up-regulated targets decrease CYP pro-tein abundance. The low abundance of CYPsdefines the hypoxic preconditioned state. With-out hypoxic preconditioning, CYPs generate ei-cosanoids, which drive the O2-ON response.By contrast, with hypoxic preconditioning or inegl-9 mutants, the CYP amounts are insufficientto generate eicosanoids and the O2-ON responseis not triggered. Neither C20 PUFAs nor over-expression of CYP-29A3 restored the defectiveO2-ON response of egl-9 mutants (figs. S9 andS10), indicating that this defect is unlikely to becaused by a general deficiency in C20 PUFAsor CYPs. Because the O2-ON response requiresEMB-8, a general CYP reductase, but only thesustained phase requires CYP-13A12, we propose
Fig. 1. n5590 suppressesthedefectofegl-9mutantsin theO2-ON response. (A)Speed graph of wild-typeanimals, showing a normalO2-ON response. Averagespeed values T 2 SEM (blue)of animals (n > 50) are shownwith step changes of O2 be-tween 20 and 0% at theindicated times. The meanspeed within 0 to 120 s afterO2 restoration is increasedrelative to that before O2restoration (P < 0.01, one-sided unpaired t test). Thedashed green line indicatesthe approximate boundary(30 s after reoxygenation)between the initial and sus-tained phases of the O2-ONresponse. (B) Speed graphof egl-9(n586)mutants, show-ing a defective O2-ON re-sponse. (C) Speed graph ofegl-9(n586); cyp-13A12(n5590)mutants, showing a restoredO2-ON response mainly inthe sustained phase (right ofthe dashed green line). Themean speed within 30 to 120 safter O2 restoration was signif-icantly higher than that ofegl-9(n586)mutants (P < 0.01).(D) Speed graph of egl-9(n586);cyp-13A12(n5590)/+ mutants,showing a restored O2-ONresponse in the sustained phase.(E) hif-1 but not cyp-13A12(n5590) suppressed the ex-pression of cysl-2::GFP by egl-9(n586) mutants. GFP fluorescence micrographs of five to seven worms aligned side by side carrying the transgene nIs470[Pcysl-2::GFP] are shown. Scale bar, 50 mm.
www.sciencemag.org SCIENCE VOL 341 2 AUGUST 2013 555
REPORTS
SNP: pkP3075 SNP: uCE3-1426
1.5 Mb
LG III
CB4856
N2
n5590 (GOF)
gk733685 (LOF)
100 bp
Time (min)
Ani
mal
spe
ed (
µm/s
ec)
egl-9(n586); nEx [cyp-13A12(+)]
Sustained phase of O2-ON response restored
P ::cyp-13A12::GFP
A B
D E
Time (min)Time (min)
Ani
mal
spe
ed (
µm/s
ec) egl-9(n586); cyp-13A12(gk733685)
O2-ON response suppressed
C
Ani
mal
spe
ed (
µm/s
ec) egl-9(n586); cyp-13A12(gk733685)/+
egl-9(n586)
O2-ON response suppressed
Wild type Wild type+ hypoxia
Time (min)
Ani
mal
spe
ed (
µm/s
ec)
F
Frac
tion
of G
FP
+ a
nim
alsegl-9(n586); cyp-13A12(n5590);
cyp-13A12(RNAi)
O2-ON response suppressed
egl-9(n586)hif-1(ia4)
* *
cyp-13A12P ::cyp-13A12(n5590)::GFP
cyp-13A12
Fig. 2. n5590 is a gain-of-function allele of cyp-13A12. (A) Geneticmapping positioned n5590 between the SNPs pkP3075 and uCE3-1426.Solid gray lines indicate genomic regions for which recombinants exhibiteda defective O2-ON response, thus excluding n5590 from those regions. Thelocations of n5590 and gk733685 are indicated in the gene diagram ofcyp-13A12. (B) Speed graph of egl-9(n586); cyp-13A12(gk733685)/+ ani-mals, showing a defective O2-ON response. (C) Speed graph of egl-9(n586);
cyp-13A12(gk733685) mutants, showing a defective O2-ON response. (D)Speed graph of egl-9(n586); nEx [cyp-13A12(+)] animals, showing a restoredO2-ON response in the sustained phase (right of the dashed green line). (E)Speed graph of egl-9(n586); cyp-13A12(n5590); cyp-13A12(RNAi) animals,showing a suppressed O2-ON response. (F) Fractions of animals expressingCYP-13A12::GFP or CYP-13A12(n5590)::GFP [*P < 0.01, two-way analysis ofvariance (ANOVA) with Bonferroni test, n = 4].
Fig. 3. Requirement of CYP-13A12 for a normal O2-ONresponse. (A) Speed graph ofcyp-13A12(gk733685) loss-of-function mutants, showing anO2-ON response with a normalinitial phase but a diminishedsustained phase (left and right,respectively, of the dashed greenline). (B) Speed graph of cyp-13A12(gk733685) mutants witha rescuing wild-type cyp-13A12transgene, showing the O2-ONresponse with a normal initialphase and sustained phase. Themean speed within 30 to 120 safter O2 restoration was higherthan that of cyp-13A12(gk733685)mutants (P < 0.01, one-sidedunpaired t test, n > 50). (C) Speedgraph of emb-8(hc69) mutantsgrown at the permissive temper-ature of 15°C with simultaneousE. coli–feeding RNAi against emb-8,showing a normal O2-ON response.(D) Speed graph of emb-8(hc69)mutants grown post-embryonicallyat the restrictive temperature of25°C with simultaneous E. coli–feeding RNAi against emb-8, show-ing a reduced O2-ON response.
C D
A B
Time (min)
Ani
mal
spe
ed (µ
m/s
ec)
O2-ON response reduced
Time (min)
Ani
mal
spe
ed (µ
m/s
ec)
emb-8(hc69) + emb-8 RNAiat 15°C
emb-8(hc69) + emb-8 RNAiat 25°CO2-ON response
No sustained phase
Sustained phase
Time (min)
Ani
mal
spe
ed (µ
m/s
ec)
cyp-13A12(gk733685)nEx [cyp-13A12(+)]
Time (min)
Ani
mal
spe
ed (µ
m/s
ec)
cyp-13A12(gk733685)
2 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org556
REPORTS
that CYP-13A12 and other CYPs act as acuteO2 sensors and produce eicosanoids, which areshort-lived and act locally (22) during reoxygen-ation to signal nearby sensory circuits that drivethe O2-ON response.
In humans, a low uptake of PUFAs or animbalanced ratio of w3-to-w6 PUFAs is associ-ated with elevated risk of stroke, cardiovasculardisease, and cancer (21, 23, 29, 30). CytochromeP450s and eicosanoid production also have beenimplicated in mammalian ischemia-reperfusion(15, 21). Nonetheless, little is known concerningthe causal relationships among and mechanismsrelating O2 and PUFA homeostasis, CYP, and
PUFA-mediated cell signaling and organismalsusceptibility to oxidative disorders. Our resultsidentify a pathway in which EGL-9–HIF-1 regu-lates CYP-eicosanoid signaling, demonstrate thatPUFAs confer a rapid response to reoxygenationvia CYP-generated eicosanoids, and provide di-rect causal links among CYPs, PUFA-derivedeicosanoids, and an animal behavioral responseto reoxygenation. Because the molecular mech-anisms of O2 and PUFA homeostasis are fun-damentally similar and evolutionarily conservedbetween nematodes and mammals (7, 11, 26),we suggest that the C. elegans O2-ON responseis analogous to themammalian tissue and cellular
response to ischemia-reperfusion injury and thatthe molecular pathway including EGL-9–HIF-1and CYPs in controlling responses to reoxygenationafter anoxia is evolutionarily conserved.
References and Notes1. A. S. Go et al., Circulation 127, e6–e245 (2013).2. H. K. Eltzschig, T. Eckle, Nat. Med. 17, 1391–1401 (2011).3. A. C. Epstein et al., Cell 107, 43–54 (2001).4. W. G. Kaelin Jr., P. J. Ratcliffe, Mol. Cell 30, 393–402
(2008).5. C. Trent, N. Tsuing, H. R. Horvitz, Genetics 104, 619–647
(1983).6. G. L. Semenza, Cell 148, 399–408 (2012).7. J. A. Powell-Coffman, Trends Endocrinol. Metab. 21,
435–440 (2010).
F
B
C
A D
E
Free
eic
osan
oids
(n
g/m
g pr
otei
ns)
17.2±1.4 4.4±0.11.8±0.6
AA (20:4 n-6)COOH
OH
19-HETE
p<0.01
Wild type
19-HETE17,18-EEQ
egl-9(n586) egl-9(n586);cyp-13A12(n5590)
Wild type egl-9(n586) egl-9(n586);cyp-13A12(n5590)
COOH
O
17,18-EEQCOOH
EPA (20:5 n-3)COOH
p<0.01
G H
I J
Cou
nts
(a.u
.)
4.0
6.0
10.0 10.4
egl-9(n586)8.0
Normal behavioral state Behavioral state in egl-9 mutants or wild-type animals after hypoxic preconditioning
Wild type
10.8 11.2 11.6
Cou
nts
(a.u
.)
4.0
6.0
8.0
Cou
nts
(a.u
.)
4.0
6.0
8.0
egl-9(n586);cyp-13A12(n5590)
17,18-EEQ 216
17,18-EEQ 53
17,18-EEQ 0
21 eicosanoidsoverall (ng/mg)
O2-ON response suppressed
Time (min)
Ani
mal
spe
ed (
µm/s
ec) fat-2(wa17) No PUFA
Ani
mal
spe
ed (
µm/s
ec) fat-2(wa17) + C20 PUFA (AA)
O2-ON response restored
Time (min)
Acquisition time (min)
FAT-2 PUFA Eicosanoid O2-ON response
CYP-13A12 and other CYPs2acute O -sensor
HIF-1HIF-1target
FAT-2 PUFA
EGL-9
Eicosanoid O2-ON response
CYP-13A12 and other CYPs
chronic O -sensor2EGL-9
HIF-1HIF-1target
chronic O -sensor2
2acute O -sensor
Fig. 4. Modulationofeicosanoidcon-centrations by EGL-9 and CYP-13A12.(A) Overall levels of free eicosanoids,calculated by adding the values of theprofiled 21 eicosanoids in the wild typeand in the egl-9(n586) and egl-9(n586);cyp-13A12(n5590) strains. (B) Schematicillustrating the conversion of arachidonicacid (AA) to 19-HETE and of eicosapen-taenoic acid (EPA) to 17,18-EEQ by CYPs.(C) Quantification of 19-HETE and 17,18-EEQ concentrations in the wild typeand in egl-9(n586); cyp-13A12(n5590)and egl-9(n586) mutant strains. Amountsof free (membrane-unbound) forms of17,18-EEQ and 19-HETE from extractsof age-synchronized young adult her-maphrodites are shown (P < 0.01, two-way ANOVA post hoc test, n = 3). Errorbars are SEM. (D to F) RepresentativeHPLC-MS traces indicating free 17,18-EEQ levels based on the spectrogramsof three MS samples: (D) wild type,(E) egl-9(n586), and (F) egl-9(n586);cyp-13A12(n5590). Peaks of 17,18-EEQat its transition m/z (mass-to-charge ratio) were measured and extracted(MassHunter). The x axis shows the retention time (minutes); the y axisshows the abundance (counts), with specific integral values over indi-vidual peaks indicated above each peak. (G) Speed graph of fat-2 mu-tants, showing a defective O2-ON response. Animals were supplementedwith the solvents used in (H) as a control. (H) Speed graph of fat-2 mu-
tants, showing the O2-ON response rescued by C20 PUFA (AA) supple-mentation. (I and J) Model of how EGL-9 and CYPs control the O2-ONresponse under (I) normoxic conditions and (J) conditions of hypoxic pre-conditioning or in egl-9 mutants (see text for details). Light blue indicateslow protein activity, low amounts of eicosanoids, or a defective O2-ONresponse.
www.sciencemag.org SCIENCE VOL 341 2 AUGUST 2013 557
REPORTS
8. G. L. Semenza, Biochim. Biophys. Acta 1813,1263–1268 (2011).
9. J. Aragonés et al., Nat. Genet. 40, 170–180 (2008).10. D. K. Ma, R. Vozdek, N. Bhatla, H. R. Horvitz, Neuron 73,
925–940 (2012).11. D. K. Ma, N. Ringstad, Front. Biol. 7, 246–253 (2012).12. M. W. Budde, M. B. Roth, Genetics 189, 521–532
(2011).13. D. R. Nelson et al., Pharmacogenetics 6, 1–42
(1996).14. O. Gotoh, Mol. Biol. Evol. 15, 1447–1459
(1998).15. R. A. Gottlieb, Arch. Biochem. Biophys. 420, 262–267
(2003).16. M. Kosel et al., Biochem. J. 435, 689–700 (2011).17. J. Kulas, C. Schmidt, M. Rothe, W. H. Schunck, R. Menzel,
Arch. Biochem. Biophys. 472, 65–75 (2008).18. B. S. Berlett, E. R. Stadtman, J. Biol. Chem. 272,
20313–20316 (1997).19. R. C. Zangar, D. R. Davydov, S. Verma, Toxicol. Appl.
Pharmacol. 199, 316–331 (2004).
20. C. A. Rappleye, A. Tagawa, N. Le Bot, J. AhringerR. V. Aroian, BMC Dev. Biol. 3, 8 (2003).
21. J. Szefel et al., Curr. Mol. Med. 11, 13–25 (2011).22. M. P. Wymann, R. Schneiter, Nat. Rev. Mol. Cell Biol. 9,
162–176 (2008).23. R. S. Chapkin, W. Kim, J. R. Lupton, D. N. McMurray,
24. D. Panigrahy, A. Kaipainen, E. R. Greene, S. Huang,Cancer Metastasis Rev. 29, 723–735 (2010).
25. C. Arnold et al., J. Biol. Chem. 285, 32720–32733(2010).
26. J. L. Watts, Trends Endocrinol. Metab. 20, 58–65(2009).
27. D. R. Harder et al., Circ. Res. 79, 54–61 (1996).28. J. P. Ward, Biochim. Biophys. Acta 1777, 1–14
(2008).29. M. Gerber, Br. J. Nutr. 107 (suppl. 2), S228–S239
(2012).30. D. Mozaffarian, J. H. Wu, J. Am. Coll. Cardiol. 58,
2047–2067 (2011).
Acknowledgments: We thank C. Bargmann, A. Fire, A. Hart,Y. Iino, J. Powell-Coffman, and C. Rongo for reagents andthe Caenorhabditis Genetics Center and the Million MutationProject for strains. H.R.H. is an Investigator of the HowardHughes Medical Institute. Supported by NIH grant GM24663(H.R.H), German Research Foundation grant ME2056/3-1(R.M.), a NSF Graduate Research Fellowship (N.B.), the MITUndergraduate Research Opportunities Program (S.Z.), anda Helen Hay Whitney Foundation postdoctoral fellowship (D.K.M.).
Supplementary Materialswww.sciencemag.org/cgi/content/full/science.1235753/DC1Materials and MethodsSupplementary TextFigs. S1 to S11Table S1References (31–70)
28 January 2013; accepted 10 June 2013Published online 27 June 2013;10.1126/science.1235753
Robust transmission of information despite the presence of variation is a fundamental problemin cellular functions. However, the capability and characteristics of information transmission insignaling pathways remain poorly understood. We describe robustness and compensation ofinformation transmission of signaling pathways at the cell population level. We calculated themutual information transmitted through signaling pathways for the growth factor–mediatedgene expression. Growth factors appeared to carry only information sufficient for a binarydecision. Information transmission was generally more robust than average signal intensitydespite pharmacological perturbations, and compensation of information transmission occurred.Information transmission to the biological output of neurite extension appeared robust. Cells mayuse information entropy as information so that messages can be robustly transmitted despitevariation in molecular activities among individual cells.
Signaling pathways transmit signals fromgrowth factors to downstream gene ex-pression, influencing various cell fate de-
cisions such as cell differentiation (1). To controlcellular responses by stimulation intensity, sig-naling pathways must reliably transmit stimula-tion intensity through their signaling activities.The reliability of signal transmission dependson the balance between signal intensity and var-iation. The smaller the signal variation, the moreinformation can be transmitted through a path-way with the same dynamic range of signal in-
tensity. Even high-intensity signals cannot bereliably transmitted if the variation in signalintensity is large. In contrast, even signals withlow intensity can be reliably transmitted if thevariation in signal intensity is small (Fig. 1A).Thus, the reliability of signal transmission de-pends on both average (mean) intensity and var-iation. As a consequence, the number of controllablestates of cellular responses is determined by thenumber of reliably transmitted signals. Intui-tively, the larger the number of reliably trans-mitted signals, the more information the signalpathway can transmit. If cellular signaling path-ways are treated as communication channels inthe framework of Shannon’s information theory(2–12), the amount of information that can bereliably transmitted through a cellular signalingpathway can bemeasured bymutual information,which corresponds to the logarithm of the av-erage number of controllable states of a cellularresponse that can be defined by varied upstreamsignals (13–15).
We evaluated the information transmissionfrom growth factors to the immediate early genes(IEGs) through various signaling pathways inPC12 cells. Nerve growth factor (NGF), pituitary
1Department of Biophysics and Biochemistry, Graduate Schoolof Science, University of Tokyo, Bunkyo-ku, Tokyo 113-0033,Japan. 2Department of Computational Biology, GraduateSchool of Frontier Sciences, University of Tokyo, Bunkyo-ku,Tokyo 113-0033, Japan. 3CREST, Japan Science and Technol-ogy Corporation, Bunkyo-ku, Tokyo 113-0033, Japan.
Fig. 1. Information transmission of signalingpathways. (A) Reliability of information transmis-sion depends on both signal intensity and varia-tion. More information can be transmitted with thesame dynamic range of signal intensity if signal var-iation is smaller. Dots denote intensities of pERKsin individual cells, and lines denote the averageintensity of pERKs. p(pERKs|NGF) denotes the distri-bution (a normalized histogram) of pERKs for agiven dose of NGF. (B) Signaling pathways fromgrowth factors, such as NGF, PACAP, and PMA tothe IEGs, such as c-FOS and EGR1. Solid lines in-dicate the reported pathways for each growth fac-tor, and gray dashed lines indicate other possiblepathways. The colored boxes are the measured mol-ecules, and white ovals are unmeasured moleculesin this study.
2 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org558
